I really like mailbag columns from my favourite writers. Take a smart, opinionated writer, add clever questions from smart readers, shake and enjoy. I don’t know where we’ll find the smart writer for this one, but you guys had plenty of good questions this week, so let’s dive right in.

mailbag_q1

*For those thrown off by the word “dive,” we’ll talk about Luis Suarez later in the column.

There are a couple of parts to this question that I feel deserve some attention, so bear with me.

First, defense is not sexy. If you are writing about football stats in public and working on defense, no one is going to read your stuff. It can be pretty disheartening. Thus very little work has been done by the analytics community in this area, and what has been done usually has a tight team or player focus  (excepting Paul Riley’s or Colin’s GK work, both of which look at GK vs. expected goals based on shot location).

However, defense is damned important. Possibly more important than offensive work overall, especially when it comes to teams that are mid-table or below and have to worry about relegation.

My own complaints stem more from the fact that I can’t quantify defense with public information. Goalkeeper stats basically don’t exist in a crunchable form, and we know they are directly connected to how a team’s defense plays. Defender stats are mostly rate stats (Ints, clearances, blocks, tackles per game). Error stats – which are enormously important because of how big a probability they have in producing opposing goals – also do not exist in a crunchable format, so we know next to nothing about them. They also don’t have much history behind them, so analysis would end up drawing conclusions from annoyingly small sample sizes.

Does this information exist? In a word, yes.

Opta, who are certainly the gold standard in football data and rapidly approaching a monopoly in this area, track all of this information. Unfortunately for people like myself, Ben, and Colin Trainor, as well as many other amateur statistical analysts, there isn’t any sort of research or student option to subscribe to Opta’s data at all. None of us are professors, but what we do is similar to work you would see done by academics in this area.

There was one pay source for error stats (and a lot more) for the Premier League last season at EPL Index, but that disappeared as a public source with the new season.  I was willing to pay 40-50£ a year for EPL Index, and I’d certainly be willing to do that per league for the rest of the leagues Opta tracks. On the other hand, I can’t pay £5000 per league per year or whatever it is they charge at the corporate and team level, so I just work with public sources. (And now we’re back to where we started.)

If Opta were to make some sort of researcher subscription model available, they’d have a new revenue source, and would dramatically speed up the research for what we know about football. That said, I’m not sure whether that fits into their business model at all. Then again, given that their position as a data provider and their desire for more and more uses for their stats to come out, maybe it’s something they should think about?

(Note: StatsBomb is not monetized, and none of us work for companies using the data.)

Here’s another important thing to realize: I spend twice as much time collecting data as I do analysing or writing it and so do most of the guys who do football analytics. It’s a minimum of about 8 hours a week, every week, just spent typing numbers into a spreadsheet. Imagine if you could wipe out 90% of that data time and put it toward research. One year from when I started looking at this stuff, we still mostly exist in the dark ages. Oh well.

Back to topic at hand, we could certainly do a lot more on the defensive side with even basic data. The big breakthrough, however, would be defensive positional information. I wrote a bit about this in the summer, but it’s become especially apparent with the release of information about NBA’s SportVu player tracking that this is a) inevitable and b) a HUGE next step in football.

All of the expected goal models out there do a lot of hand waving about defensive positioning.  “These shots occurred from these spots and are typically this probability of scoring a goal.” There’s nothing wrong with it and it’s done by necessity, but having positioning information for every single player on the pitch would change everything.

The following questions were produced in about sixty seconds of thought on what this information would let us answer.

  • Who shoots the best under defensive pressure?
  • How close do defenders need to be to shooters to have the best chance of disrupting or blocking a shot?
  • Is it better to shoot from 22 yards with 3 yards of space or 20 yards with 1.5 yards of space?
  • Which GK are just inherently positioned wrong on long shots?
  • Where SHOULD GK be positioned on long shots to have the best chance of saving shots from locations X, Y, and Z?

The questions are endless, and despite the fact that many of them look like research for pure knowledge, the vast majority have practical applications. Do the research, find the results, create the teaching points for your coaches. This is true at the league level and at the national federation level. It feels like this is something that would immediately get funded by the Germany Ministry of Sport and the US Soccer Federation, but would get argued about for years and never funded in England.

Hell, forget Opta (who barely broke even during their last fiscal year), and think about the three biggest monetary entities in the sport. Isn’t this the sort of greater good project – one that benefits the teams, the players, and the fans – that should see some of the billions in revenue that FIFA, UEFA, and the Premier League generate every year thrown at it? Learning more about the sport as a whole can only be a positive, right?

So yeah. Make more info public, and we can do a lot better with basic defensive info. Make a lot more info available to the right people and you’ll completely change what we know about the game of football.

mailbag_q2

At first, I got whatever stats I could from whatever source I could. These days there are four primary locations I get information from, all powered by Opta.

The first one on my list is WhoScored. I like their layout, how quickly the site responds, and the fact that they have information going back to the start of the 2009-10 season for most of the bigger leagues. It would significantly improve my life if they would opt to post information in a Per90 format instead of how they do it, and there’s more basic information I would like to see (tackle success rates and defensive errors plus minutes played are probably tops on the list), but overall I really like the site.

For team stats, if they made some really simple, useful additions like goals against per game, goals for per game, and shots on target conceded per game, I’d probably kiss Cris Acconci directly on the mouth when I eventually meet him.

Also, if anybody ever crosses the information at WS plus the breakdowns for performance by player over at Transfrmarket, plus shot location info like Colin does, it would turn into the site I use 90% of the time instead of 60%.

The second site I use regularly for stats is Squawka. Squawka has some incredibly powerful information and breakdowns, displays some simply lovely graphical representations, and wraps everything in an interface I hate. I assume they do this so that script kiddies can’t scrape the everloving shit out of their website, which I completely understand, but it interferes with my enjoyment of the site. It is, however, the only place you can get some extremely important information like passing breakdowns and percentages for players by type.

Without the information there, it probably would have taken much longer for the world to prove it is a helluva lot harder to pass the ball with your head than with your feet.

Example: Olivier Giroud has a passing accuracy of 78% with his feet this season (182/232)and 32.5% with his head (29/89).

And here’s a second example, just because the mental image made me laugh.

Example 2: Xavi had 2405 successful passes with his feet last season out of 2507 total “general” passes (95.9%). He also had 11 successful and 5 missed headed passes (69%).

Conclusion: Using your head is hard, even for Xavi. (And Xavi likely never gets beat up on aerial challenges the way forwards do.)

The third place I get stats is from the FourFourTwo StatsZone, which is kind of the gold standard for graphical stats displays in football. It’s really easy to plonk images created there into article pieces, and they have some great information.  They’ve also recently added at least one league I can’t find anywhere else, which guarantees them traffic from me.

The final place I get data from is The Football App on my ipad.  I’m apparently not the only one, since The Guardian actually did a profile on them last month, and they’ve had explosive growth. Transcribing all of their information from ipad to PC is about as much fun as going to the dentist, but the app itself is really quite good.

(Note: Apparently it’s available to people who run Windows 8.1 or higher. This would require me choosing to use Windows 8. Having used Windows 8, I fully intend to skip this generation of MS upgrades and hope the next version is less awful.)

mailbag_q3

Despite being asked by a reader named UnfitForPurpose, this is a great question because it lends itself to a couple of different analytical approaches.

If you are looking purely at money, then Bayern Munichare far and away the best team in Germany, and have the single biggest monetary advantage of any team in Europe versus the rest of the teams in their country. They have been one of the best teams in Europe consistently over the last five years (at least) while making a profit.

In fact, they have made a profit twenty years in a row.

Mind. Blown.

Incredibly, they have sandbagged their spending overall to make certain that the Bundesliga remains consistently competitive. The problem they have is that they also want to win the Champions League, and other teams across Europe don’t share their cares about competitive league balance. Is Bayern choosing to splash out more of their revenue to win more CL medals going to cause issues in the coming seasons for the rest of the Bundesliga? It certainly could. The Pep Guardiolas of the world don’t come cheap.

For the moment, Dortmund are every bit as good as Bayern and now have enough spending power to stay fairly close in player quality in the first XI. The problem is that Bayern are literally two-deep with exceptional talent at every position. Dortmund go about 15 deep and then fill in with a bunch of potentially talented kids. There’s a large separation between those two and the rest of the league, but at least Bayern have competitive company for now.

If you want simple league metrics, Spain remains the most ridiculous. Atletico are really exceptional this season, but Barcelona and Real Madrid always win the league, and the TV distro looks like this:

la_liga_distro

Teams on the bottom get no money. Goal differences for the top 2 teams are regularly absurd. With Barce and Madrid getting nearly 12x the TV revenue alone compared to the smallest clubs, and 3X the third and fourth closest, competitive balance will be a problem in Spain forever and ever and ever and ever…

In terms of what the predictive model thinks, SerieA now has the biggest separation between how good the team at the top is versus all the other teams. Since Totti and Gervinho went down injured, Juventus have started playing much better, and Roma have fallen off a bit. This has lead to a .75 goal expectation for Juve versus the second place team (Roma) on a neutral pitch, which is currently the largest of any league I track in Europe. I don’t think the gulf is nearly that large if Roma is fully healthy, but Juve are both good and deep. It will be interesting to see if the other top teams in Italy can keep up with them, or whether The Old Lady from Turin will gradually pull away from what is actually a pretty good league this season.

mailbag_q4

It’s only six matches, but the man with the Cameron avatar is exactly correct – Suarez has been out of his mind this season.

suarez_sixgames

A goal a game in the Premier League is awesome. Suarez is currently bettering that by a third. Multiply that by a full 38 game season and you end up with 50 goals. You know who has posted rate stats to get 50 league goals in a season?

Ronaldo.

Messi.

That’s it, at least in modern times. That’s bonkers.

It’s not just the goal rate that’s great though – Suarez is incredibly involved in the Liverpool offense. 41 passes per game looks closer to a midfielder than a forward, and his key pass and dribble stats are strong. The Loss of Possession stats are still a bit too high for comfort (Offsides, Dispossession, Turnovers), but look what you get for it! Suarez has been amazing.

I wrote something this summer on how Suarez was incredibly inefficient, to the point of driving fans and analysts (and presumably Brendan Rodgers) crazy. His conversion rate over the last two seasons was the lowest of any major forward in Europe. That plus the discipline issues completely drove me off him as a player, and I believed that Liverpool should absolutely snap off a bid of £40M from interested teams. Except maybe from Arsenal, who they would be directly competing with to make the Champions League next year.

HOWEVER… there was this nagging issue of his performance for Uruguay clogging up the analysis. When playing for his national team, Suarez wasn’t good plus inefficient – he was actually just great. If Liverpool could ever figure out how to translate Uruguayan Luis into Liverpudlian Luis, watch out. I didn’t think it could happen.

So far this season, Suarez is converting shots to goals at a 22% clip and getting 42% of his shots on target. That’s light years better than his recent Premier League conversion (38.5% SOT last year and 12.3% non-penalty conversion rate).

It’s just six matches so far this year, but maybe Rodgers and co. did it? If so, I am happy to own up on being wrong. If nothing else, they should now be able to hold out for a bid more like what PSG paid for Cavani’s services next summer, should they even want to sell/Suarez still wants to leave.

Can it continue?

This one is a bit like Arsenal winning the Premier League for me. Historic precedent and underlying stats suggest that their hot run won’t continue for the rest of the season, but it’s in the realm of possibility now. And if Suarez averages a goal a game for Liverpool for the rest of the season, and if they manage to keep the rest of their attack healthy…

Liverpool are definitely going to make the Champions League. They might even battle it out for the league title.

Those are still pretty big ifs, but they no longer reside in the realm of wishful thinking.

What’s up with Crystal Palace’s manager situation?

Okay, I made this one up because I wanted to talk about it. It’s a strange spot for a number of reasons, not least because it seems to be taking quite a while to hire a manager in the most desirable league in the world.  The issues become a lot more obvious when you work through the problem.

Right now Crystal Palace are probably 95% likely to be relegated. They don’t have a ton of talent, they sold one of their two best attackers from last season to Manchester United, and their other one (Glen Murray) is still recovering from a blown ACL. That’s not a great place to start a battle for Premier League survival.

Unlike Sunderland, who hired Gus Poyet earlier this month while also navigating the bottom of the Premier League, Palace’s fundamentals just aren’t as good. They don’t have as much money, and their roster isn’t nearly as talented as the Wearside club. Palace do have a good youth program in place though, and they are clearly thinking in the right direction regarding future player recruitment by hiring Moody as Sporting Director.

Now say you are one of the better managers in the English Championship right now. Your goal is to get to the Premier League.

Why wouldn’t you take the Palace job?

Because there’s a second unstated half of that goal – “To make it to the Premier League… and stay there.” If I’m an up and coming manager, I don’t want to just make it to the PL, I also want a fighting chance to stay there, and Palace’s current situation doesn’t provide that. If I believe in myself and my team in the lower league, it’s actually more likely that I will be in the Premier League next year if I get my current team promoted instead.

Similar issues arise with foreign managers. They look at the squad composition and how unlikely it is they will survive this season and say, “Meh, I have it better here. Something better will come up.”

I don’t actually know what Palace are likely to do, but I think they have two real options. They should either find a caretaker to see out this season, (and pay them a massive bonus if they were to somehow accomplish a miracle and survive), and then do their entire search again in the summer when their league status is known.

Or they should hire a younger manager from the lower leagues (League One and below) who can develop talent to finish this season and set up their run for the Championship next season (where they will suddenly be a club flush with cash). All their other options just have too many incentives vs. interest issues to end up hiring a manager you actually want running the club for the long term.

That’s it for me today. Thank you for all of the questions you guys submitted, and maybe we’ll do this again toward Christmas, when half of the leagues will be on their winter breaks.

–@mixedknuts

 

 

  • http://Mitch.com Mitch

    As a biased Spurs fan, I thought Real Madrid would’ve better spent their money by buying Suarez for half of what they spent on Bale. I’m glad to see his performance this year are proving me right. However, I also hope he drops off a little bit so we don’t get gazumped again for a CL spot.

  • Toshack

    Quality work as ever Ted.
    Speaking of defence. If “The Numbers game” is to be believed the expected points gained is higher when conceding zero goals compared to if you score one goal. Way higher. 0>1.
    So yeah, defence is important…
    And Error stats. In defence it’s what is NOT happening that matters. No errors (or no tackles for that matter). Take tackle stats. Yes. It’s like Xabi Alonso remarked. Tackling is the last resort only when all other options are out. Maldini was one of the best defenders ever (mentored by Baresi). He almost never got his white shorts dirty. He was just “there” to fend off the attack.
    As they say in the book: It’s the dog that didn’t bark that matters in defence…
    So yeah, we need lots of stats, but different kinds of stats compared to the stats tracking the goal scoring giants.
    Cheers,
    Peter

  • marlon

    There’d be a lot more defensive analysis out there if the same metrics were available for defence as they are for attack (SOT conceded etc.).

    One stat I think might be a decent proxy for defensive pressure/positioning is shots conceded/final third pass conceded. Next to useless on an individual level but could show some interesting systems stuff.

    I know analysts seem to jealously guard their hard collected data, but maybe if we shared our spreadsheets then we could get more done?

    • http://BenPugsley Ben Pugsley

      I have the data for a shots/final third pass conceded. I’ll publish it before this weekend. One thought: it’s probably better to strip out set-piece shots I guess?

      • marlon

        Thanks. Probably should’ve mentioned how much data you publish amongst all the complaining there…

        Yeah, stripping out set pieces would be better if possible.

  • Pingback: Find the leak! Analyzing PSV’s midfield… | 11tegen11()

Improve Performance and Productivity in Your Club:
State-of-the-art Football Analytics